Re: [PATCH 50/51] core: changed retval of _cairo_composite_rectangles_init_for_mask() to cairo_bool_t

On 21.12.2015 13:41, Chris Wilson wrote:

Hi,

> Honestly, I prefer the interface to be consistent (i.e. all the
> functions for the operations return the same type), consistent in error
> propagation, and returning the status allows for more flexibilty in
> future.

My intention is to limit the range to the actual domain.
If we just use the 'big' status enum everywhere, we (or the compiler)
don't really know which return values could happen.

> At the moment the callsites are essentially:
> 
> 	test; cjmp; undo stack; ret
> 
> afterwards they become;
> 
> 	test; cjmp; undo stack; mov NOTHING_TO_DO, eax; ret
> 
> or more likely;
> 	test; cjmp -> ret
> becomes
> 	test; cjmp; mov NOTHING_TO_DO, eax; jmp -> ret

OTOH, the caller might be more complex - it needs to explicitly check
for NOTHING_TO_DO. I'd suspect a check for zero/nonzero is a bit faster
than test for specific value.

The current implementation explicitly assigns the return value to a
local (likely on-stack) variable an then checks it. At least we should
directly compare the it, w/o assignment, IMHO.

--mtx

--
Enrico Weigelt,
metux IT consulting
+49-151-27565287
-- 
cairo mailing list
[email protected]
http://lists.cairographics.org/mailman/listinfo/cairo